DongbaMIE: A Multimodal Information Extraction Dataset for Evaluating Semantic Understanding of Dongba Pictograms

📖arXiv | 🤗Dataset

Timeline

📢 [2025-03-06] DongbaMIE dataset released.

📢 [2025-03-05] Paper and repo released.

Constructing DongbaMIE Dataset

Project Overview

This repository contains the following files:

• generate_data_qwen2vl.py for building qwen2-vl format data
• gpt4o_vqa_test.py for getting gpt4o and gemini test results
• metric.py for performance evaluation

Semantic visualization result

DongbaMIE dataset statistics

Result

Overall Performance

The results of extracting the four semantic dimensions of objects, actions, relations and attributes simultaneously in a single reasoning

Citation

If you find our project useful, please consider citing:

@inproceedings{bi-etal-2025-dongbamie,
    title = "{D}ongba{MIE}: A Multimodal Information Extraction Dataset for Evaluating Semantic Understanding of Dongba Pictograms",
    author = "Bi, Xiaojun  and
      Li, Shuo  and
      Xing, Junyao  and
      Wang, Ziyue  and
      Luo, Fuwen  and
      Qiao, Weizheng  and
      Han, Lu  and
      Sun, Ziwei  and
      Li, Peng  and
      Liu, Yang",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.findings-emnlp.51/",
    doi = "10.18653/v1/2025.findings-emnlp.51",
    pages = "976--990",
    ISBN = "979-8-89176-335-7"
}